The global memory shortage linked to rising AI demand could force a large number of consumer electronics companies out of business by the end of 2026, according to comments attributed to Pua Khein-Seng, CEO of Phison.

In a summary posted on X by user 駿HaYaO, Pua reportedly said that many system vendors may go bankrupt or exit product lines due to limited memory supply. The original interview was conducted in Chinese and is available on YouTube without English subtitles, so the remarks cannot be independently verified verbatim.

《獨家專訪 #群聯 #Phison 執行長潘健成》
核心結論
1. 記憶體供需嚴重失衡將持續到2030年
AI為剛需，DRAM與NAND Flash極度缺貨。原廠要求預付3年貨款（電子業史無前例），賣方市場空前強勢。原廠內部預估缺到2030年，甚至10年不見盡頭。

2. 消費電子將大量死亡… https://t.co/FvVnb8tT2I

— 駿HaYaO (@QQ_Timmy) February 14, 2026

Production Cuts Expected

According to the translated summary, Pua warned that consumer electronics will experience widespread failures from late this year through 2026. He reportedly said mobile phone production could decline by 200 to 250 million units, while PC and television output would also fall significantly.

The shortage has already driven sharp increases in memory prices, complicating the production of various computing-related devices.

NVIDIA’s Rubin GPUs to Blame

Pua also reportedly addressed the impact of Nvidia’s next-generation Rubin AI GPUs. If tens of millions of Vera Rubin units ship, each requiring more than 20TB of SSD storage, that would consume roughly 20 percent of last year’s global NAND production capacity, excluding additional data storage demand, according to the summary.

Long-Term Shortage Looms

The summary further claims that memory manufacturers are now requesting three years of prepayment, a practice described as unprecedented in the electronics sector. Manufacturers are also said to estimate that the shortage could last until 2030 or potentially another decade.

Major memory producers, including Samsung, Micron, SK Hynix, Kioxia, and Yangtze Memory, have announced investments in new capacity. However, bringing new facilities online typically requires at least two years from announcement to production, and equipment remains in high demand.

China’s additional output is expected to account for only 3 to 5 percent of global capacity in its early stages, which would not close a projected 10 to 20 percent supply gap. Domestic demand within China is also described as strong, limiting the likelihood of surplus exports.

Sustainability Debate

The X user suggested that extended product lifespans and increased repairs could be a potential upside of constrained supply. However, overall electronics production continues to grow, with output shifting toward energy-intensive AI servers and GPUs rather than consumer hardware.

Snapchat announced it will roll out creator subscriptions, introducing a feature similar to offerings already available on Instagram, Facebook, and OnlyFans.

Under the new system, subscribers will gain access to exclusive content shared through Snaps and Stories. They will also receive priority replies from creators and view Stories without ads.

Subscription Pricing and Access

Creators will be able to set their own monthly subscription prices. These prices can be aligned with Snapchat’s recommended pricing tiers.

The feature will first become available to select US-based creators starting February 23. In the US, iOS users will be able to subscribe to participating creators’ accounts.

Snapchat said it plans to expand the subscription option to Canada, France, and the UK in the coming weeks.

Mustafa Suleyman, CEO of Microsoft AI at Microsoft, said in a recent interview with the Financial Times that artificial intelligence could automate most professional computer-based tasks within the next 12 to 18 months.

Suleyman predicted human-level performance on most professional work, including accounting, legal services, marketing, and project management. He said jobs that involve sitting at a computer are especially vulnerable. His comments align with recent warnings from AI researcher Matt Shumer and Sam Altman of OpenAI, who have both described rapid AI progress as potentially disruptive.

Earlier this year, Dario Amodei of Anthropic warned that AI could eliminate half of entry-level white-collar roles. Jim Farley of Ford Motor Company said AI might cut US white-collar jobs in half. At the World Economic Forum in Davos, Elon Musk of SpaceX said artificial general intelligence could arrive as early as this year.

Current Impact Remains Limited

Despite these forecasts, AI adoption in professional services has so far produced modest results. A 2025 report from Thomson Reuters found lawyers, accountants, and auditors using AI for targeted tasks such as document review and routine analysis. Productivity gains were described as marginal and did not indicate widespread job displacement.

In some cases, AI has reduced efficiency. A study from nonprofit Model Evaluation and Threat Research found that software developers took 20% longer to complete tasks when using AI tools.

Economic gains appear concentrated in the technology sector. Research from Torsten Slok, chief economist at Apollo Global Management, showed Big Tech profit margins rising more than 20 percent in the fourth quarter of 2025. By contrast, the broader Bloomberg 500 Index showed little change. Slok also cited Wall Street expectations for the S and P 500, noting investors do not anticipate higher earnings from AI outside the tech sector.

Job Cuts

There are early signs of workforce impact. Employment consultancy Challenger, Gray, and Christmas reported about 55,000 AI-related job cuts in 2025. Microsoft eliminated 15,000 roles last year, although it did not attribute the reductions directly to AI. In a July memo, Microsoft CEO Satya Nadella said the company needed to reimagine its mission for a new era.

Markets have reacted sharply to AI developments. Software stocks recently declined amid fears of automation in the software as a service sector, a selloff some analysts labeled the SaaSpocalypse. The downturn followed announcements from Anthropic and OpenAI about agentic AI systems designed to handle functions traditionally performed by SaaS companies.